Biological activity of substrate-bound basic fibroblast growth factor ...

Oncogene (2002) 21, 3889 ± 3897
ã 2002 Nature Publishing Group All rights reserved 0950 ± 9232/02 $25.00
www.nature.com/onc
Biological activity of substrate-bound basic ®broblast growth factor
(FGF2): recruitment of FGF receptor-1 in endothelial cell adhesion contacts
Elena Tanghetti 1 , Roberto Ria 2 , Patrizia Dell'Era 1 , Chiara Urbinati 1 , Marco Rusnati 1 ,
Maria Grazia Ennas 3 and Marco Presta* ,1
1
Unit of General Pathology and Immunology, Department of Biomedical Sciences and Biotechnology, School of Medicine,
University of Brescia, Italy; 2 Department of Biomedical Sciences and Human Oncology, School of Medicine, University of Bari,
Italy; 3 Department of Cytomorphology, University of Cagliari, Italy
Substrate-bound FGF2 promotes endothelial cell adhesion
by interacting with a v b 3 integrin. Here, endothelial
GM7373 cells spread and organize focal adhesion plaques
on immobilized FGF2, ®bronectin (FN), and vitronectin
(VN). a v b 3 integrin, paxillin, focal adhesion kinase,
vinculin and pp60 src localize in cell-substratum contact
sites on FGF2, FN or VN. However, only immobilized
FGF2 induces a long-lasting activation of extracellular
signal-regulated kinases 1/2 (ERK 1/2 ) and cell proliferation
that was inhibited by the ERK 1/2 inhibitor PD 098059 and
the tyrosine kinase (TK) inhibitor tyrphostin 23, pointing
to the engagement of FGF receptor (FGFR) at the basal
side of the cell. To assess this hypothesis, GM7373 cells
were transfected with a dominant negative TK 7 -DFGFR1
mutant (GM7373-DFGFR1 cells) or with the full-length
receptor (GM7373-FGFR1 cells). Both transfectants
adhere and spread on FGF2 but GM7373-DFGFR1 cells
do not proliferate. Also, parental and GM7373-FGFR1
cells, but not GM7373-DFGFR1 cells, undergo morphological
changes and increased motility on FGF2-coated
plastic. Finally, FGFR1, but not TK 7 -DFGFR1, localizes
in cell adhesion contacts on immobilized FGF2. In
conclusion, substrate-bound FGF2 induces endothelial
cell proliferation, motility, and the recruitment of FGFR1
in cell-substratum contacts. This may contribute to the
cross talk among intracellular signaling pathways
activated by FGFR1 and a v b 3 integrin in endothelial cells.
Oncogene (2002) 21, 3889 ± 3897. DOI: 10.1038/sj/
onc/1205407
Keywords: cell adhesion; cell proliferation; angiogenesis;
integrin; FGF; tyrosine kinase; FGF receptor
Introduction
*Correspondence: M Presta, General Pathology, Dept. Biomedical
Sciences and Biotechnology, Via Valsabbina 19, 25123 Brescia, Italy;
E-mail: presta@med.unibs.it
Received 21 June 2001; revised 8 February 2002; accepted 19
February 2002
Angiogenesis is a multi-step process that plays a key
role in di€erent physiological and pathological conditions,
including embryonic development, wound repair,
in¯ammation, and tumor growth (Carmeliet and Jain,
2000). It begins with the degradation of the basement
membrane by activated endothelial cells that will
migrate and proliferate, leading to the formation of
solid endothelial cell sprouts into the stromal space.
Then, vascular loops are formed and capillary tubes
develop with deposition of new basement membrane
and accessory cell recruitment (Carmeliet, 2000). A
close interaction exists among cell-adhesive proteins of
the extracellular matrix (ECM), their integrin receptors,
and soluble angiogenesis growth factors during
each step of the angiogenesis process (Ingber and
Folkman, 1989a,b; Davis et al., 1993; Brooks et al.,
1994; Plopper et al., 1995).
Basic ®broblast growth factor (FGF2) is one of the
best characterized modulators of angiogenesis. FGF2
induces neovascularization in vivo in di€erent experimental
models (Basilico and Moscatelli, 1992) and is
implicated in the growth of new blood vessels during
wound healing and chick embryo development (Broadley
et al., 1989; Ribatti et al., 1995). In vitro, FGF2
induces cell proliferation, migration, and production of
proteases in endothelial cells (Moscatelli et al., 1986)
by interacting with speci®c tyrosine-kinase (TK)
receptors (FGFRs) and with heparan sulfate proteoglycans
(HSPGs) of the cell surface (Johnson and
Williams, 1993). Also, FGF2 modulates integrin
expression in endothelium (Enenstein et al., 1992;
Klein et al., 1993).
Integrins are a family of transmembrane, heterodimeric
adhesion receptors comprised of a and b
subunits. The combination of di€erent subunits
originates distinct integrin molecules that mediate cell
adhesion to a variety of adhesive proteins of the ECM
such as ®bronectin (FN), vitronectin (VN), thrombospondin,
laminin and collagens (Albelda and Buck,
1990; Hynes, 1992). Besides mediating cell adhesion,
the interaction of integrins with cell-adhesive proteins
plays a crucial role in regulating the response of
endothelial cells to soluble growth factors, including
FGF2 (Ingber and Folkman, 1989a,b; Stromblad and
Cheresh, 1996). Also, a v b 3 integrin is highly expressed
by endothelial cells during angiogenesis and is required
to sustain neovascularization induced by FGF2

3890
Biological activity of immobilized FGF2
E Tanghetti et al
(Brooks et al., 1994; Friedlander et al., 1995; Eliceiri et
al., 1998). Interestingly, FGFRs, integrins, and intracellular
transducers may co-localize in focal adhesion
contacts (Plopper et al., 1995, Miyamoto et al.,
1996). Despite these observations, the molecular
mechanism(s) underlying the relationship between the
FGF2/FGFR system and the cell adhesion machinery
are not fully elucidated.
Large amounts of FGF2 are present in ECM both in
vivo and in vitro (Vlodavsky et al., 1987; Folkman et
al., 1988). Collagen-bound FGF2 is mitogenically
active in situ for BALB/c-3T3 ®broblasts (Smith et
al., 1982) and FGF2 immobilized onto heparin-coated
surfaces promotes endothelial cell adhesion (Baird et
al., 1988) and PC12 cell adhesion and di€erentiation
(Schubert et al., 1987). Thus, ECM-bound FGF2 may
induce endothelial cell adhesion and act at the same
time as a localized, persistent stimulus for angiogenesis
by interacting with di€erent cell-surface molecules.
Indeed, previous data obtained in our laboratory had
shown that FGF2 interacts also with a v b 3 integrin and
that this interaction mediates the capacity of the
angiogenic growth factor to induce cell adhesion,
mitogenesis, and urokinase-type plasminogen activator
upregulation in endothelial cells (Rusnati et al., 1997).
Interestingly, the endothelial cell-adhesive capacity of
FGF2 does not require the interaction of the growth
factor with FGFRs and/or HSPGs that are instead
able to mediate a FGF2-dependent cell-cell interaction
via the formation of a ternary FGFR/FGF2/HSPG
complex (Richard et al., 1995).
In the present study we investigated the molecular
mechanisms mediating the mitogenic activity of
substrate-bound FGF2 in endothelial cells. The results
demonstrate that immobilized FGF2 induces focal
adhesion plaque formation and the activation of
extracellular signal-regulated kinases (ERK 1/2 ) in
bovine aortic endothelial GM7373 cells. This is
paralleled by a signi®cant increase in the proliferation
rate of adherent cells. GM7373 cells transfected with a
dominant negative, truncated TK 7 FGFR1 mutant
adhere and spread but have lost the capacity to
proliferate on immobilized FGF2. Intact FGFR1, but
not the truncated receptor, localizes in the focal
adhesion contacts induced by FGF2. These data shed
a new light on the cross-talk between endothelial cell
adhesion and proliferative events during angiogenesis.
and the localization of vinculin (Figure 1B) and
paxillin (not shown) demonstrate that immobilized
FGF2 induces the formation of focal adhesion contacts
in adherent GM7373 cells. Similar results were
obtained for cells adherent to FN or VN (data not
shown). Under the same experimental conditions, GM
7373 cells attach (Figure 1A) but do not spread (not
shown) on polylysine (PLL)-coated plastic.
To con®rm these observations, GM7373 cells were
seeded on FGF2, FN, VN, or PLL, allowed to adhere
for 6 h, and stripped by PBS/EDTA washes (Culp,
1976; Del Rosso et al., 1992). Cell-substratum contact
sites, that represent 3 ± 4% of the surface membrane of
monolayered cells (Del Rosso et al., 1992), were then
extracted and analyzed by Western blotting. As shown
in Figure 2, FGF2-adherent plasma membrane remnants
contain paxillin, a v b 3 integrin, focal adhesion
kinase (FAK), and pp60 src . Similar results were
obtained with cells adherent to FN and VN but not
with cells adherent to PLL.
Taken together the results demonstrate that immobilized
FGF2 promotes focal adhesion plaque formation
in endothelial cells with the recruitment of signal
transducing molecules including FAK and pp60 src .
Immobilized FGF2 induces endothelial cell proliferation
Next, we evaluated the capacity of plastic-bound FGF2
to promote cell proliferation in endothelial GM 7373
cells. To this purpose, cells were allowed to adhere for
2 h on di€erent substrata, including native or heatinactivated
FGF2, FN, or VN. Then, non-adherent
cells were removed and adherent cells were incubated
in low serum. After a 24 h incubation, cells were
trypsinized and counted. As shown in Figure 3, only
immobilized native FGF2 is able to exert a rapid and
signi®cant mitogenic response in adherent cells that
Results
Substrate-bound FGF-2 promotes focal adhesion plaque
formation in endothelial cells
Previous observations had shown that substrate-bound
FGF2 promotes endothelial cell adhesion via a v b 3
integrin engagement (Rusnati et al., 1997). Accordingly,
fetal bovine aortic endothelial GM 7373 cells
adhere onto non-tissue culture plates coated with
native or heat-inactivated FGF2, FN, or VN (Figure
1A). Also, the distribution and organization of F-actin
Figure 1 GM 7373 cell adhesion to FGF2-coated plastic. (A)
Non tissue culture plastic plates were incubated with carbonate
bu€er containing 20 mg/ml of BSA, FGF2, heat-inactivated FGF2
(hi-FGF2), FN, VN, or PLL. GM 7373 cells were seeded onto
coated plates and allowed to adhere for 2 h at 378C. Then, the
number of adherent cells was evaluated. Each point is the
mean+s.e.m. of three determinations in duplicate. (B) GM 7373
cells adherent to immobilized FGF2 were stained with rodaminate-phalloidin
(a) or anti-vinculin antibody (b)
Oncogene

continue to proliferate for further 24 h. In contrast,
FN adherent cell cultures showed a signi®cant increase
in cell number only 72 h after seeding (Figure 3B). No
proliferation was observed for cells seeded on plastic
coated with heat-inactivated FGF2 (not shown) or
BSA. It must be pointed out that no signi®cant
di€erences in the levels of vascular endothelial growth
factor (ranging between 16 and 30 pg/ml) were detected
by ELISA in the conditioned medium of GM 7373
cells grown on the di€erent substrata.
Downstream signaling triggered by the binding of
FGF2 to its TK + FGFRs encompasses the activation
of mitogen-activated protein kinase kinase (MEK) with
consequent phosphorylation of ERKs (Giuliani et al.,
1999). Accordingly, a slow but long-lasting increase in
Biological activity of immobilized FGF2
E Tanghetti et al
ERK 1/2
phosphorylation was observed in GM7373
cells seeded on immobilized FGF2 but not on
immobilized FN (Figure 4A). Indeed, integrin engagement
by FN is known to cause a rapid but transient
activation of this signaling pathway (Miyamoto et al.,
1996). The MEK inhibitor PD 098059 (Alessi et al.,
1995) prevented ERK 1/2 activation whereas SB 210313,
a selective inhibitor of p38 kinase (Cuenda et al., 1995),
was ine€ective (Figure 4B). Accordingly, PD 098059
inhibited the proliferation of GM7373 cells adherent to
FGF2-coated plastic whereas SB 210313 was ine€ective
(Figure 4D). The mitogenic response triggered by
immobilized FGF2 was inhibited also by the TK
inhibitor tyrphostin 23 (Boyer and Thiery, 1993), but
not by tyrphostin 63 (Figure 4D) here used as a
negative control (Gazit et al., 1989). None of the
compounds tested was able to a€ect the adhesion of
GM7373 cells to immobilized FGF2 (Figure 4C).
3891
Figure 2 Western blot analysis of cell-substratum contact sites.
GM 7373 cells were seeded onto plates coated with FGF2, FN,
VN, or PLL and allowed to adhere for 6 h at 378C. Then cells
were detached from the plastic with 3 mM EDTA/PBS washes.
Plasma membrane remnants were washed three times with PBS
and extracted. Aliquots (30 mg) of the extracted material were
analysed by Western blotting with the indicated antibodies
respect to cells adherent at T 0 population doublings in respect to cells adherent at T 0
Figure 4 ERK 1/2 phosphorylation by immobilized FGF2. GM
7373 cells were seeded onto FN- or FGF2-coated plastic. Western
blot analysis of the cell extracts was performed 1 and 2 h after
seeding using anti-phospho-ERK 1/2 antibodies (A). In B, cells
were seeded on FGF2-coated plastic in the absence or in the
presence of 50 mM SB 210313 (SB) or PD 098059 (PD). Western
blot analysis of the cell extracts was performed 2 h after seeding
using anti-phospho-ERK 1/2 antibodies. In parallel experiments,
Figure 3 Mitogenic activity of immobilized FGF2. GM 7373 GM 7373 cells were seeded onto FGF2-coated plates in the
cells were seeded onto plates coated with FGF2, heat-inactivated
FGF2 (hi-FGF2), FN, or VN and allowed to adhere for 2 h at
378C (T 0 ). Then, non-adherent cells were removed and adherent
cells were incubated in fresh medium containing 0.4% FCS. Cells
were trypsinized and counted 24 h (A) or 24, 48 and 72 h (B) after
seeding. Data represent the mean+s.e.m. of three determinations
in duplicate and are expressed as cell population doublings in
absence or in the presence of 100 mM tyrphostin 23 (t23), 100 mM
tyrphostin 63 (t63), 50 mM SB 210313 (SB) or 50 mM PD 098059
(PD) and allowed to adhere for 2 h at 378C (T 0 ). Then, nonadherent
cells were removed and adherent cells were counted
immediately (C) or after a 24 h-incubation in fresh medium
containing 0.4% FCS (D). Data represent the mean+s.e.m. of
three determinations in duplicate. In D, data are expressed as cell
Oncogene

3892
Biological activity of immobilized FGF2
E Tanghetti et al
Taken together, the data indicate that substratebound
FGF2 retains its mitogenic capacity that
requires TK activity and ERK 1/2 phosphorylation.
These observations raise the possibility that FGFR
localized at the basal side of endothelial cells is
involved in mediating the mitogenic response of
adherent cells to the immobilized growth factor.
Figure 5 Expression of dominant negative DFGFR1 in GM
7373 cells. GM 7373 cells were transfected with a retroviral
expression vector harboring the full length TK + FGFR1 cDNA
or the truncated TK 7 FGFR1 cDNA. Stable transfectants were
isolated, generating GM7373-FGFR1 and GM7373-DFGFR1
cells, respectively. Then, binding of 125 I-FGF2 to high anity
receptors was evaluated as described in Materials and methods
(A). Also, cell extracts were probed with anti-FGFR1 antibodies
by Western blotting (B). In C, parental, GM7373-FGFR1, and
GM7373-DFGFR1 cells were incubated for 20 min in the
presence of the indicated concentrations of soluble FGF2. Then,
Western blot analysis of the cell extracts was performed using
anti-phospho-ERK 1/2 antibodies. a, parental GM 7373 cells; b,
GM7373-DFGFR1 cells; c, GM7373-FGFR1 cells
Dominant negative TK FGFR1 abolishes the response of
endothelial cells to immobilized FGF2
To assess the role of FGFR in transducing a mitogenic
signal in endothelial cells adherent to immobilized
FGF2, GM 7373 cells were transfected with a mutated
FGFR1 cDNA carrying a stop codon in the
juxtamembrane domain. These cells, named GM7373-
DFGFR1 cells, will express a dominant negative,
truncated TK 7 DFGFR1 devoid of its TK domain
and C-terminus (Li et al., 1994). In parallel, distinct
GM7373 cell cultures were transfected with the full
length TK + FGFR1 cDNA, thus generating GM7373-
FGFR1 cells.
As shown in Figure 5A, both GM7373-DFGFR1 and
GM7373-FGFR1 cells binds 125 I-FGF2 with a capacity
signi®cantly higher than that of parental cells. Western
blot analysis of the cell extracts probed with a
monoclonal anti-FGFR1 antibody evidenced the presence
of a Mr 130 000 immunoreactive band in the cell
extract of GM7373-FGFR1 cells (Figure 5B), corresponding
to the full length receptor and comigrating
with a fainter band present in the extract of parental
cells. An intense Mr 90 000 immunoreative band,
corresponding to the overexpressed truncated receptor,
was instead detected in the cell extract of GM7373-
DFGFR1 cells (Figure 5B). A signi®cant increase of
ERK 1/2 phosphorylation was detectable in parental and
GM7373-FGFR1 cells adherent to tissue culture plastic
and treated with soluble FGF2. No ERK 1/2 phosphorylation
was instead detected in FGF2-treated GM7373-
DFGFR1 cells, thus con®rming the dominant negative
e€ect of the truncated TK 7 receptor (Figure 5C).
GM7373-DFGFR1 and GM7373-FGFR1 cells adhere
to immobilized FGF2, FN, or VN with an
eciency similar to that shown by parental cells
(Figure 6A). Also, their capacity to adhere to
immobilized FGF2 was prevented by the highly speci®c
monoclonal LM 609 antibody directed to a v b 3
(Cheresh, 1987) (Figure 6A). Accordingly, all the cell
lines express similar amounts of a v b 3 , as evidenced by
Western blot analysis of the cell extracts (data not
shown). However, GM7373-DFGFR1 cells have lost
the ability to proliferate when seeded on FGF2-coated
plastic; this ability is instead retained by GM7373-
FGFR1 transfectants (Figure 6B). As observed for
parental cells, GM7373-FGFR1 and GM7373-
DFGFR1 cells do not proliferate when seeded on
heat-inactivated FGF2, FN, or VN. It must be pointed
out that no signi®cant di€erences were observed in the
proliferation rate of the three cell lines when seeded on
tissue culture plastic and maintained in 10% FCS (data
not shown).
Administration of FGF2 to the culture medium
a€ects the appearance of endothelial cells that acquire
an elongated, ®broblast-like morphology associated to
increased cell motility (Tsuboi et al., 1990). Similarly,
GM7373-FGFR1 cells adherent to FGF2-coated
plastic took an elongated appearance with a crisscross
pattern within 24 ± 48 h after seeding (Figure 7a,b).
Also, crawling cells characterized by lamellipodia and
microspikes at the leading edge were frequently
observed (Figure 7c ± e). Similar even though less
dramatic changes were observed 48 ± 72 h after seeding
in parental GM 7373 cells adherent to immobilized
FGF2 (data not shown), possibly re¯ecting the lower
number of FGFR receptors expressed by parental cells
in respect to the transfectants. In contrast, no
morphological changes were observed in FGF2-
adherent GM7373-DFGFR1 cells that retained a ¯atter
cobblestone-like appearance throughout the whole
experimental period (Figure 7f,g). Speci®city of the
e€ect was demonstrated by the lack of activity of
immobilized FN (Figure 7h ± l) and VN (not shown)
that did not in¯uence the morphological features of
adherent GM7373-FGFR1 and parental cells.
Immobilized FGF2 recruits FGFR1 in cell-substratum
contact sites
The above data suggest that immobilized FGF2 can
interact with FGFR1 at the basal side of the adherent
Oncogene

Biological activity of immobilized FGF2
E Tanghetti et al
3893
Figure 6 E€ect of dominant negative DFGFR1 on the mitogenic activity of immobilized FGF2. Parental, GM7373-FGFR1, and
GM7373-DFGFR1 cells were seeded onto plastic coated with BSA, FGF2 (in the absence or in the presence of monoclonal LM609
anti-a v b 3 antibody), heat-inactivated FGF2, FN, or VN and allowed to adhere for 2 h at 378C (T 0 ). Then, non-adherent cells were
removed and adherent cells were counted immediately (A) or after a 24 h-incubation in fresh medium containing 0.4% FCS (B).
Data represent the mean+s.e.m. of three determinations in duplicate. In B, data are expressed as cell population doublings in
respect to cells adherent at T 0
cells, thus triggering a mitogenic and morphogenic
response. Accordingly, FGFR1 and paxillin co-localize
in GM 7373 cells adherent to immobilized FGF2
(Figure 8A). Also, Western blot analysis of cellsubstratum
contact sites organized by GM7373-
FGFR1 transfectants adherent to FGF2-coated plastic
demonstrates the presence of FGFR1 in this plasma
membrane fraction (Figure 8B). Semi-quantitative
Western blot analysis indicates that more than 50%
of the total amount of FGFR1 accumulates in this
fraction. No FGFR1 was detected in contacts formed
by GM7373-FGFR1 cells adherent to FN.
To con®rm that the interaction with immobilized
FGF2 causes a redistribution of FGFR1 at the basal
side of the cell, GM7373-FGFR1 cells were seeded in
96-well plates at 75 000 cells/cm 2 on tissue culture
plastic or on plastic coated with FGF2 or FN. After
6 h, adherent cell monolayers were incubated for 2 h at
48C with 125 I-FGF2 (10 ng/ml) and its binding to high
anity FGFRs was evaluated (Rusnati et al., 1996).
FGF2-adherent cells showed a signi®cant reduction in
the capacity to bind 125 I-FGF2 at the apical side of the
cell monolayer (5+2 c.p.m./well) when compared to
cells adherent to FN (151+20 c.p.m./well) or to tissue
culture plastic (225+30 c.p.m./well) (n=3). This occurs
in the absence of signi®cant di€erences in the total
amount of FGFR1 expressed by these cells under the
various experimental conditions, as shown by Western
blot analysis of the cell extracts with anti-FGFR1
antibodies (data not shown).
No truncated DFGFR1 accumulates in contacts
organized by GM7373-DFGFR1 adherent to immobilized
FGF2 (Figure 8b) despite the high levels of
expression of the receptor in these cells (see Figure 5).
These data indicate that the recruitment of FGFR1 by
immobilized FGF2 is speci®c and that the interaction
of FGF2 with the extracellular domain of the receptor
is not sucient for its cooption in the adhesion
contacts. Accordingly, the TK inhibitor tyrphostin 23
prevented the recruitment of intact FGFR1 and caused
a decrease in the amount of pp60 src in focal contacts of
GM7373-FGFR1 transfectants adherent to immobilized
FGF2. No e€ect was instead exerted by
tyrphostin 63 (Figure 9).
Taken together the data indicate that immobilized
FGF2 is able to recruit FGFR1 in cell adhesion
contacts of adherent GM7373 cells. The intracellular
domain and/or the TK activity of the receptor are
required for its cooption in these structures.
Discussion
Immobilized FGF2 interacts with a v b 3 integrin and
promotes endothelial cell adhesion and spreading
(Rusnati et al., 1997). Our results demonstrate that
substrate-bound FGF2 retains its biological activity,
stimulating ERK 1/2 phosphorylation, cell proliferation,
and motility in adherent endothelial GM7373 cells.
This is paralleled by the recruitment of FGFR1 in cellsubstratum
contact sites. Overexpression of the
dominant negative TK 7 DFGFR1, treatment with the
TK inhibitor tyrphostin 23, or treatment with the
MEK inhibitor PD 098059 inhibit the mitogenic
activity of immobilized FGF2 without a€ecting a v b 3 -
mediated cell adhesion.
ECM may act as a physiological reservoir for
extracellular FGF2 (Vlodavsky et al., 1987; Folkman
et al., 1988). Various enzymes, including heparanase,
thrombin, collagenases, plasmin, and urokinase-type
Oncogene

3894
Biological activity of immobilized FGF2
E Tanghetti et al
Figure 8 FGFR1 recruitment in cell-substratum contact sites.
(A) GM 7373 cells adherent to immobilized FGF2 were
immunostained with anti-FGFR1 (A) and anti-paxillin (B)
antibodies. FGFR1 co-localizes in paxillin-positive focal adhesion
contacts (arrowheads). (B) GM7373-FGFR1 and GM7373-
DFGFR1 cells were seeded onto plates coated with FGF2, FN,
or PLL and allowed to adhere for 6 h at 378C. Then cells were
detached from the plastic with 3 mM EDTA/PBS washes. Plasma
membrane remnants were washed three times with PBS and
extracted. Aliquots (30 mg) of the extracted material were
analysed by Western blotting with antibodies directed against
vinculin or FGFR1. The anticipated position corresponding to
DFGFR1 migration is also indicated
Figure 7 Morphology of GM7373-FGFR1 cells adherent to
immobilized FGF2. GM7373-FGFR1 cells (a ± e, h ± l) and
GM7373-DFGFR1 cells (f, g) were allowed to adhere onto glass
coverslips coated with 20 mg/ml of FGF2 (a±g)orFN(i, l). After
48 h, cells were ®xed and photographed under a phase contrast
inverted microscope at 406 magni®cation (a, f, h), processed for
scanning electron microscopy and photographed at 8006 (b),
3006 (c, g), 20006 (d), or 6006 (i) magni®cation, or stained
with rodaminate-phalloidin and photographed at 6306 magni®cation
(e, l). Note the elongated shape and crisscross pattern of
GM7373-FGFR1 cells adherent to FGF2 (a, b, e) when compared
to the ¯atter and more regular appearance of FN-adherent cells
(h±l) and of FGF2-adherent GM7373-DFGFR1 cells (f, g). In c,
GM7373-FGFR1 cells migrate onto the FGF2-coated substratum
(white boxed enlarged in c showing a crawling cell characterized
by lamellipodia and microspikes at the leading edge, also evident
in e after rodaminate-phalloidin staining)
plasminogen activator release ECM-bound FGF2
(Ribatti et al., 1999, and references therein) suggesting
that the balance between storage and release of
FGF2 in ECM, as well as the integrity of the matrix,
may regulate the biological e€ects of this growth
factor on endothelium. On the other hand, collagenbound
FGF2 is mitogenically active in situ for
BALB/c-3T3 ®broblasts (Smith et al., 1982) and
FGF2 immobilized onto heparin-coated surfaces
promotes endothelial cell adhesion (Baird et al.,
1988) and PC12 cell adhesion and di€erentiation
(Schubert et al., 1987). Our ®ndings extend these
Figure 9 TK activity is required for FGFR1 recruitment in cellsubstratum
contact sites. GM7373-FGFR1 cells were seeded onto
FGF2-coated plates and allowed to adhere for 6 h at 378C in the
absence or in the presence of 100 mM tyrphostin 63 (t63) or
tyrphostin 23 (t23). Then cells were detached from the plastic with
3mM EDTA/PBS washes. Plasma membrane remnants were
washed three times with PBS and extracted. Aliquots (30 mg) of
the extracted material were analysed by Western blotting with
antibodies directed against the indicated proteins
observations and demonstrate that endothelial cells
adherent to FGF2 proliferate and acquire a migra-
Oncogene

Biological activity of immobilized FGF2
E Tanghetti et al
tory phenotype. FGF2 bound to plastic resists to
extraction with urea, methanol, or ethanol and it is
removed only by drastic treatment with detergents
like Triton X-100 or SDS (Smith et al., 1982;
Rusnati et al., 1997). Accordingly, no FGF2
internalization was observed in cells adherent to
FGF2-coated plastic in which 125 I-FGF2 was used
as a tracer (M Rusnati, unpublished observations).
These data indicate that FGF2 may represent an
angiogenic stimulus also when tightly bound to the
substratum.
Under our experimental conditions the amount of
FGF2 bound to non-tissue culture plastic corresponds
to 10 12 molecules/cm 2 (Rusnati et al., 1997). Thus, each
endothelial cell adheres onto approximately 10 7
molecules of immobilized FGF2. This causes the
a v b 3 -mediated formation of focal adhesion contacts
and actin cytoskeletal organization. The analysis of the
components of the cell-substratum contact sites indicates
that immobilized FGF2 activates the all series
of speci®c stages of hierarchies of protein interactions
that occur during integrin response, as observed for
typical cell-adhesion proteins like FN and VN
(Miyamoto et al., 1995). Indeed, the presence of a v b 3
integrin, FAK, vinculin, pp60 src , and paxillin in cellsubstratum
contact sites demonstrate that immobilized
FGF2 is able to cause integrin receptor aggregation,
integrin occupancy, cytoplasmic tyrosine phosphorylation,
and actin cytoskeletal organization (Miyamoto et
al., 1995).
As stated above, endothelial cell adhesion to
immobilized FGF2 is followed by a rapid and
signi®cant increase in cell proliferation. However,
integrin engagement is not sucient per seÁ to trigger
a mitogenic response. Indeed, GM 7373 cells adhere
and spread also on heat-inactivated FGF2, FN, or VN
leading to the formation of focal adhesion contacts in
the absence of a signi®cant increase of their rate of
proliferation that occurs only at late time points.
Conversely, overexpression of the dominant negative
DFGFR1, tyrphostin 23, and PD 098059 completely
abolish the mitogenic response to immobilized FGF2
without a€ecting a v b 3 -mediated cell adhesion. Moreover,
neutralizing anti-a v b 3 monoclonal and polyclonal
antibodies do inhibit cell proliferation induced by
soluble FGF2 in GM 7373 cells grown on tissue
culture plastic (Rusnati et al., 1997). Thus, interaction
of FGF2 with a v b 3 integrin is necessary but not
sucient to transduce a mitogenic signal that requires
the activation of a functional TK + FGFR.
Previous observations had shown that integrin
clustering at sites of contact of FN-coated beads with
®broblast cell surface is accompanied by a transient
accumulation of various growth factor TK receptors,
including FGFR1 (Miyamoto et al., 1996). Here we
show that immobilized FGF2, but not immobilized
FN or PLL, triggers a long-lasting accumulation of
FGFR1 at the cell-substratum contact sites of
adherent GM7373-FGFR1 transfectants. Six hours
after adhesion, cell-substratum contact sites, that
represent 3 ± 4% of the surface membrane of monolayered
cells (Del Rosso et al., 1992), contain more
than 50% of total FGFR1 molecules, clearly indicating
that the interaction with immobilized FGF2
concentrates the receptor in these cell membrane
remnants. Accordingly, FGF2-adherent cells showed
a dramatic reduction in the number of high anity
125
I-FGF2 binding sites present on the apical side of
the cell monolayer when compared to cells adherent to
FN or to tissue culture plastic, thus con®rming that
interaction with immobilized FGF2 causes a redistribution
of FGFRs at the basal side of the cell.
Interestingly, the truncated receptor overexpressed by
GM7373-DFGFR1 cells does not accumulate in
FGF2-induced contact sites, despite its ability to bind
FGF2 in a manner undistinguishable from the wildtype
receptor. Thus, the interaction of the extracellular
domain of the receptor with the immobilized growth
factor is not sucient to guarantee a long-lasting
accumulation of FGFR1 in the focal adhesion
contacts that may require the activation of the
intracellular TK moiety of the receptor and/or its
interaction with other intracellular proteins. The
ability of tyrphostin 23 to prevent FGFR1 recruitment
and the capacity of FGFR1 to associate with pp60 src
and cortactin (Zhan et al., 1994) support this
hypothesis. Recently, the vascular endothelial growth
factor receptor VEGFR2/KDR has been shown to
interact directly with a v b 3 integrin (Soldi et al., 1999).
By using the same experimental approaches, we have
been unable to observe a direct interaction between
FGFR1 and a v b 3 integrin (E Tanghetti, unpublished
data). The dissection of the molecular basis of
FGFR1 recruitment in cell-substratum contact sites
by immobilized FGF2 will require further investigation.
Previous observations had implicated ERK activation
in FGF2 signaling (Besser et al., 1995; Giuliani et
al., 1999) and angiogenesis (Eliceiri et al., 1998). FGF2
causes a long-lasting phosphorylation of ERK 1/2 in
parental and GM7373-FGFR1 cells, but not in
GM7373-DFGFR1 transfectants. Also, the MEK
inhibitor PD 098059 prevents the mitogenic response
of parental and GM7373-FGFR1 cells to immobilized
FGF2. These data emphasize the role of ERK 1/2 in
signal transduction activated by FGFR1 occupancy in
endothelium.
Several experimental evidences support the hypothesis
that integrins collaborate with TK receptors in
transducing the intracellular signals triggered by
growth factors in target cells (Miyamoto, 1995, 1996;
Kumar, 1998; Giancotti and Rouslathi, 1999). We
report here that immobilized FGF2 interacts with a v b 3
integrin and FGFR1, a€ecting di€erent aspects of the
angiogenic phenotype of endothelial cells, including cell
adhesion, proliferation, morphogenesis, and motility.
Our data indicate that FGFR1 and a v b 3 integrin may
be favored in their cross-talk by the long-lasting
structural vicinity that occur at the basal aspect of
the endothelium where they colocalize in cell-substratum
contact sites after adhesion onto immobilized
FGF2.
3895
Oncogene

3896
Materials and methods
Biological activity of immobilized FGF2
E Tanghetti et al
Materials
Human recombinant FGF2 was expressed and puri®ed from
transformed E. coli cells as described (Isacchi et al., 1991).
Anti-a v b 3 monoclonal LM 609 antibody was from Chemicon
International (Temecula, CA, USA). For Western blotting,
monoclonal anti-FGFR1 antibody and polyclonal antipp60
src antibody were from Upstate Biotechnology (Lake
Placid, NY, USA) and Santa Cruz Biotechnology (Santa
Cruz, CA, USA), respectively. PD 098059 and SB 210313
were from Calbiochem (San Diego, CA, USA) and RBI
(Natick, MA, USA), respectively. PLL, tyrphostin 23,
tyrphostin 63, rodaminate-phalloidin, and anti-vinculin antibody
were from Sigma (St. Louis, MO, USA). Monoclonal
anti-FAK antibody was a gift from G Tarone (University of
Turin, Italy). Bovine FN was from Harbor Bio-Products
(Norwood, MA, USA) and human VN was from Becton
Dickinson Labware (Bedford, MA, USA).
Cell cultures and transfection
Fetal bovine aortic endothelial GM 7373 cells, corresponding
to the BFA-1c multilayered transformed clone (Grinspan et
al., 1983), were grown in Eagle's minimal essential medium
(MEM) containing 10% FCS, vitamins, essential and non
essential amino acids.
Plasmids 91023b-¯g (murine IIIc variant of FGFR-1
cDNA), pCEP4-¯g 1.2 (truncated murine IIIc variant of
FGFR-1 cDNA) and pCB7 (a plasmid that carries the neo r
gene) were provided by A Mansukhani and C Basilico (New
York University Medical Center, New York, NY, USA).
GM7373 cells were transfected with 91023b-¯g in association
(10 : 1, wt:wt) with pCB7 according to a calcium phosphate
precipitation protocol as described (Rusnati et al., 1996) and
selected with 250 mg/ml of G418 (Sigma, St. Louis, MO,
USA). Parallel cultures were transfected with pCEP4-¯g 1.2
and selected with 200 mg/ml of hygromycin B (Boehringer
Mannheim GmbH, Mannheim). For each transfection,
resistant clones were tested for 125 I-FGF2 binding capacity
(Rusnati et al., 1996) and by Western blotting using anti-
FGFR1 antibody (Upstate Biotechnology, Lake Placid, NY,
USA).
Cell adhesion assay
One hundred ml aliquots of 100 mM NaHCO 3 , pH 9.6
(carbonate bu€er), containing the adhesive molecule under
test were added to polystyrene non-tissue culture microtiter
plates. After 16 h of incubation at 48C the solution was
removed and wells were washed three times with cold
phosphate bu€ered saline (PBS). Then, wells were incubated
for 1 h at 378C with 1.0 mg/ml BSA and washed
extensively with PBS. For the cell-adhesion assay, con¯uent
cultures of GM 7373 cells were trypsinized, washed and
resuspended with the appropriate medium. Previous observations
had indicated that low concentrations of serum
were required in some experiments for optimal cell
adhesion to FGF2-coated plastic (Rusnati et al., 1997).
For this reason, 1% FCS was utilized routinely in celladhesion
experiments. Fifty thousand GM 7373 cells were
resuspended in 200 ml of medium and were immediately
seeded onto wells coated with the molecule under test.
Routinely, cell-adhesion was allowed to occur for 2 h at
378C. Then, wells were washed with 2 mM EDTA/PBS.
Adherent cells were trypsinized and counted.
Western blot analysis of cell-substratum contact sites
Cells were seeded at 75 000 cells/cm 2 on plastic coated with
the di€erent substrata. After 6 h, cells were detached from
the plastic with 3 mM EDTA/PBS (Del Rosso et al., 1992).
The material remaining on the plastic, representing plasma
membrane remnants, was washed three times with PBS and
extracted with 0.1 M Tris-HCl (pH 8.1) containing 1 mM
PMSF, 0.2% sodium-deoxycholate, 1% Triton-X100. Aliquots
(30 mg) of the extracted material were analysed by
Western blotting with the indicated antibodies.
Evaluation of the mitogenic activity of immobilized FGF2
Parental and transfected GM7373 cells were seeded at
130 000 cells/cm 2 onto 96-well plates coated with the di€erent
substrata and incubated for 2 h at 378C with MEM
containing 1% FCS. Then, cells were washed with 2 mM
EDTA/PBS (T 0 ). Adherent cells were incubated for further
24, 48 or 72 h in fresh MEM containing 0.4% FCS. In some
experiments, medium was added with di€erent inhibitors. At
the end of incubation, cells were trypsinized and counted in a
Burker chamber. Data are expressed as cell population
doublings in respect to cells adherent at T 0 .
Western blot analysis of ERK 1/2 phosphorylation
Parental and transfected GM7373 cells were seeded at 42 000
cells/cm 2 in 35 mm tissue culture dishes. After adhesion, cells
were incubated in serum-free MEM added with 10 mg/ml
transferrin and 1 mg/ml BSA. After 24 h, cells were
incubated for 20 min at 378C with no addition or 2.0 or
10.0 ng/ml FGF2. Then, Western blot analysis of the cell
extracts was performed as described (Besser et al., 1995)
using anti-phospho-ERK 1/2 antibody (New England Biolabs,
Inc., Beverly, MA, USA).
In some experiments, GM7373 cells were seeded at 90 000
cells/cm 2 in 35 mm dishes coated with FGF2 or FN in the
absence or in the presence of 50 mM PD 098059 or SB
210313. After 1 or 2 h, cell extracts were analysed for ERK 1/2
phosphorylation as above.
For all the experiments, equal loading of the di€erent lanes
of the gel were con®rmed by Western blotting of the stripped
membranes with anti-ERK 1/2 antibody (not shown).
Immunocytochemistry
Cells were seeded onto FGF2- or FN-coated glass coverslips
in MEM added with 1% FCS. After 2 ± 6 h, cells were
washed, ®xed in 3% paraformaldehyde, 2% sucrose in PBS,
permeabilized with 0.5% Triton-X100, and saturated with
3% BSA in PBS. Then, in a ®rst set of experiments, cells
were incubated with monoclonal anti-vinculin antibody
followed by ¯uorescinated anti-mouse IgG plus rodaminatephalloidin.
In a second set of experiments, cells were
incubated with monoclonal anti-paxillin antibody (Transduction
Laboratories, Lexington, KY, USA) followed by Alexa
Fluor 594 anti-mouse IgG (Molecular Probes, Eugene, OR,
USA) and polyclonal anti-FGFR1 antibody (Santa Cruz
Biotechnology) followed by biotinylated swine anti-rabbit
IgG (Dako, Glostrup, Denmark) and rhodol green conjugated
streptavidin (Molecular Probes).
Scanning electron microscopy
Glass coverslips (10 mm in diameter) were placed within
24-well tissue culture plates and coated overnight at 48C
Oncogene

with carbonate bu€er containing 20 mg/ml of FGF2, FN,
or VN. Then, plates were washed three times with cold
PBS. Cells were seeded at 130 000/cm 2 and allowed to
adhere onto glass coverslips for 2 h in MEM added with
1% FCS. Then, cells were washed with 2 mM EDTA in
PBS. Adherent cells were incubated for further 24 ± 72 h in
fresh MEM containing 0.4% FCS. Cells were then ®xed
with 1.5% glutaraldehyde, 1% paraformaldehyde in 0.15 M
cacodylate bu€er (pH 7.4) for 1 h. Coverslips were then
washed, dehydrated, critical point dried with CO 2 and
sputter coated with 3.2 nM platinum with Emitech K575
apparatus (Emitech, UK). Cells were then viewed under
Hitachi scanning electron microscope ®nd emission mod.
Biological activity of immobilized FGF2
E Tanghetti et al
S4000 (Hitachi, Japan) operated at 15 kV and photographed.
Acknowledgments
This work was supported by MURST (60% and Co®n
2000 to M Rusnati and to M Presta.; Centro di Eccellenza
`IDET' to M Presta), CNR (Progetto Finalizzato Biotecnologie),
Associazione Italiana per la Ricerca sul Cancro,
Istituto Superiore di SanitaÁ (AIDS Project), and Centro per
lo Studio del Trattamento dello Scompenso Cardiaco
(University of Brescia) to M Presta
3897
References
Albelda SM and Buck CA. (1990). FASEB J., 4, 2868 ±
2880.
Alessi DR, Cuenda A, Cohen P, Dudley DT and Saltiel AR.
(1995). J. Biol. Chem., 270, 27489 ± 27494.
Baird A, Schubert D, Ling N and Guillemin R. (1988). Proc.
Natl. Acad. Sci. USA, 85, 2324 ± 2328.
Basilico C and Moscatelli D. (1992). Adv. Cancer Res., 59,
115 ± 165.
Besser D, Presta M and Nagamine Y. (1995). Cell Growth
Di€er., 6, 1009 ± 1017.
Boyer B and Thiery JP. (1993). J. Cell Biol., 120, 767 ± 776.
Broadley KN, Aquino AM, Woodward SC, Buckley-
Sturrock A, Sato Y, Rifkin DB and Davidson JM.
(1989). Lab. Invest., 61, 571 ± 575.
Brooks PC, Richard AF and Cheresh DA. (1994). Science,
264, 569 ± 571.
Carmeliet P. (2000). Nature Med., 6, 389 ± 395.
Carmeliet P and Jain RK. (2000). Nature, 407, 249 ± 257.
Cheresh DA. (1987). Proc. Natl. Acad. Sci. USA, 84, 6471 ±
6475.
Cuenda A, Rouse J, Doza YN, Meier R, Cohen P, Gallagher
TF, Young PR and Lee JC. (1995). FEBS Lett., 364, 229 ±
233.
Culp LA. (1976). Biochemistry, 15, 4094 ± 4104.
Davis CM, Danehower SC, Laurenza A and Molony JL.
(1993). J. Cell. Biochem., 51, 206 ± 218.
Del Rosso M, Pedersen N, Fibbi G, Pucci M, Dini G,
Anichini E and Blasi F. (1992). Exp. Cell Res., 203, 427 ±
434.
Eliceiri BP, Klemke R, Stromblad S and Cheresh DA. (1998).
J. Cell Biol., 140, 1255 ± 1263.
Enenstein J, Waleh NS and Kramer RH. (1992). Exp. Cell
Res., 203, 499 ± 503.
Folkman J, Klagsbrun M, Sasse J, Wadzinsky M, Ingber D
and Vlodavsky I. (1988). Am. J. Pathol., 130, 393 ± 400.
Friedlander M, Brooks PC, Sha€er RW, Kincaid CM,
Varner JA and Cheresh DA. (1995). Science, 270, 1500 ±
1502.
Gazit A, Yaish P, Gilon C and Levitzki A. (1989). J. Biol.
Chem., 264, 14503 ± 14509.
Giancotti FG and Rouslathi E. (1999). Science, 285, 1028 ±
1032.
Giuliani R, Bastaki M, Coltrini D and Presta M. (1999). J.
Cell Sci., 112, 2597 ± 2606.
Grinspan JB, Stephen NM and Levine EM. (1983). J. Cell
Physiol., 114, 328 ± 338.
Hynes RO. (1992). Cell, 69, 11 ± 25.
Ingber DE and Folkman J. (1989a). Cell, 58, 803 ± 805.
Ingber DE and Folkman J. (1989b). J. Cell Biol., 109, 317 ±
330.
Isacchi A, Statuto M, Chiesa R, Bergonzoni L, Rusnati M,
Sarmientos P, Ragnotti G and Presta M. (1991). Proc.
Natl. Acad. Sci. USA, 88, 2628 ± 2632.
Johnson DE and Williams LT. (1993). Adv. Cancer Res., 60,
1±41.
KleinS,GiancottiFG,PrestaM,AlbeldaSM,ClaytonAB
and Rifkin DB. (1993). Mol. Biol. Cell, 4, 973 ± 982.
Kumar CC. (1998). Oncogene, 17, 1365 ± 1373.
Li Y, Basilico C and Mansukhani A. (1994). Mol. Cell Biol.,
14, 7600 ± 7669.
Miyamoto S, Teramoto H, Coso OA, Gutkind JS, Burbelo
PD, Akiyama SK and Yamada KM. (1995). J. Cell Biol.,
131, 791 ± 805.
Miyamoto S, Teramoto H, Gutkind JS and Yamada KM.
(1996). J. Cell Biol., 135, 1633 ± 1642.
Moscatelli D, Presta M and Rifkin DB. (1986). Proc. Natl.
Acad.Sci.USA,83, 2091 ± 2095.
Plopper GE, McNamee HP, Dike LE, Bojanowski K and
Ingber DE. (1995). Mol. Biol. Cell, 6, 1349 ± 1365.
Ribatti D, Urbinati C, Nico B, Rusnati M, Roncali L and
Presta M. (1995). Dev. Biol., 170, 39 ± 49.
Ribatti D, Leali D, Vacca A, Giuliani R, Gualandris A,
Roncali L, Nolli ML and Presta M. (1999). J. Cell Sci.,
112, 4213 ± 4221.
Richard C, Liuzzo JP and Moscatelli D. (1995). J. Biol.
Chem., 270, 24188 ± 24196.
RusnatiM,Dell'EraP,UrbinatiC,TanghettiE,Massardi
ML, Nagamine Y, Monti E and Presta M. (1996). Mol.
Biol. Cell, 7, 369 ± 381.
Rusnati M, Tanghetti E, Dell'Era P, Gualandris A and
Presta M. (1997). Mol. Biol. Cell, 8, 2449 ± 2461.
Schubert D, Ling N and Baird A. (1987). J. Cell Biol., 104,
635 ± 643.
Smith JC, Singh JP, Lillquist JS, Goon DS and Stiles CD.
(1982). Nature, 269, 154 ± 156.
Soldi R, Mitola S, Strasly M, De®lippi P, Tarone G and
Bussolino F. (1999). EMBO J., 18, 882 ± 892.
Stromblad C and Cheresh DA. (1996). Trends Cell Biol., 6,
462 ± 468.
Tsuboi R, Sato Y and Rifkin DB. (1990). J. Cell Biol., 110,
511 ± 517.
Vlodavsky I, Folkman J, Sullivan R, Fridman R, Ishai-
Michaeli R, Sasse J and Klagsbrun M. (1987). Proc. Natl.
Acad.Sci.USA,84, 2292 ± 2296.
Zhan X, Plourde S, Hu X, Friesel R and Maciag T. (1994). J.
Biol. Chem., 269, 1±4.
Oncogene